Fig. 6: Clustered heatmaps displaying the changes in systematic distance (sys_dist_m) and Optimal String Alignment (OSA) distances in the Shanghai dataset.

a1, a2 Changes in mean sys_dist (a1) and OSA (a2) between statistical centroid in the Shanghai local dialectal varieties (labelled on the right). b1, b2 Changes in mean sys_dist (b1) and OSA (b2) between the Shanghai Shiqu (central urban) dialect and the other Shanghai sub-dialects (labelled on the right). c1, c2 Changes in mean sys_dist (c1) and OSA (c2) between Standard Chinese (SC) and Shanghai local dialectal varieties (labelled on the right).The phonemic units of syllables (Syl), segmental combinations (Seg), onset consonants (Ons), rhymes (Rhy), vowels (Vow), final consonants (Fin), and tone classes (Ton) are labelled at the bottom of each plot, ordered according to the result of clustering. The heatmaps are colour-coded so that blue indicates decreasing distances and red indicates increasing distances, with higher saturation indicating greater changes. Significance codes and directions of paired t-tests comparing by-sub-dialect mean old and new sys_dist/OSA are annotated at the top of each plot (red↑for increasing, blue ↓for decreasing).